A Satellite Model of Forest Flammability Marc K. Steininger • Karyn Tabor • Jennifer Small • Carlos Pinto • Johan Soliz • Ezequiel Chavez Received: 15 March 2012 / Accepted: 6 May 2013 / Published online: 13 June 2013 Ó Springer Science+Business Media New York 2013 Abstract We describe a model of forest flammability, based on daily satellite observations, for national to regional applications. The model defines forest flamma- bility as the percent moisture content of fuel, in the form of litter of varying sizes on the forest floor. The model uses formulas from the US Forest Service that describe moisture exchange between fuel and the surrounding air and pre- cipitation. The model is driven by estimates of temperature, humidity, and precipitation from the moderate resolution imaging spectrometer and tropical rainfall measuring mission multi-satellite precipitation analysis. We provide model results for the southern Amazon and northern Chaco regions. We evaluate the model in a tropical forest-to- woodland gradient in lowland Bolivia. Results from the model are significantly correlated with those from the same model driven by field climate measurements. This model can be run in a near real-time mode, can be applied to other regions, and can be a cost-effective input to national fire management programs. Keywords Tropical forest  Fire risk  Drought  Remote sensing  Amazon  Bolivia Introduction Forest Flammability Models Models of forest flammability allow a better understanding of fire risk and are sought by national forest agencies to support forest management. Most models use surface-cli- mate data from either in situ measurements or nearby weather stations. The models are typically applied on a daily basis, where the previous day’s conditions are mod- ified with the current day’s climate. Examples include using the length of time since the last significant rainfall event and a bucket model of soil moisture evaporation (e.g., Nepstad and others 2004; Ray and others 2005; INPE 2011). These parameters are used as indicators of risk, based on underlying assumptions of relationships to the moisture of the available fuels and empirical relationships with the frequency of fire occurrence. Some models include seasonal forecasting, such as that of Chen and others (2011), who use the empirical relationship between sea- surface temperatures and fire activity to forecast fire-season severity, although at a very coarse 5° resolution. The USFS defines fire risk as the ‘‘chance of fire start- ing, as determined by the presence and activity of causative agents’’ (NWCG 2013). It defines flammability as the M. K. Steininger (&)  K. Tabor Conservation International, 2011 Crystal Drive, Suite 500, Arlington, VA 22202, USA e-mail: msteininger@conservation.org J. Small Department of Geography, University of Maryland at College Park, 2181 LeFrak Hall, College Park, MD 20742, USA Present Address: J. Small National Aeronautics and Space Administration Goddard Space Flight Center, Mail Code: 614.4, Greenbelt, MD 20771, USA C. Pinto  J. Soliz  E. Chavez Departamento de Geografia, Museo Noel Kempff Mercado, Universidad Autonomia Gabriel Rene Moreno, Avenida Irala 565, Box 2489, Santa Cruz de la Sierra, Bolivia Present Address: C. Pinto Fundacio ´n Amigos de la Naturaleza, Castilla Postal 2241 Santa Cruz de la Sierra, Bolivia 123 Environmental Management (2013) 52:136–150 DOI 10.1007/s00267-013-0073-1